图片转换为dataloader和数据集加载函数

img = Image.open(path)
img = img.resize((200, 100), Image.ANTIALIAS)
transf = transforms.ToTensor()
img_tensor = transf(img) 

# 第一维为batch
train_x = torch.normal()
train_y = torch.normal()
train_set = Data.TensorDataset(train_x, train_y)
train_iter = Data.DataLoader(dataset=train_set, batch_size=batch_size, shuffle=True, num_workers=0)


def load_data_fashion_mnist(batch_size, num_workers=0, resize=None):
    '''load dataset fashion_mnist'''
    trans = [transforms.ToTensor()]
    if resize:
        trans.insert(0, transforms.Resize(resize))
    trans = transforms.Compose(trans)
    mnist_train = torchvision.datasets.FashionMNIST(root="../data",
                                                    train=True,
                                                    transform=trans,
                                                    download=True)
    mnist_test = torchvision.datasets.FashionMNIST(root="../data",
                                                   train=False,
                                                   transform=trans,
                                                   download=True)
    return (torch.utils.data.DataLoader(mnist_train, batch_size, shuffle=True, num_workers=num_workers),
            torch.utils.data.DataLoader(mnist_test, batch_size, shuffle=False, num_workers=num_workers))